View Bug Details

IDProjectCategoryView StatusLast Update
0002580DCP-o-maticBugspublic2023-09-21 00:08
Reporteroverlookmotel Assigned Tocarl  
Status feedbackResolutionopen 
PlatformMacOSOS XOS Version10.14
Product Version2.16.59 
Target Version2.16.66 
Summary0002580: Inaccurate MainSoundConfiguration in metadata

I've noticed what I think is inaccurate data in the SMPTE CompositionMetadataAsset in CPL of DCPs created with 2.16.x.

Input: SMPTE DCP made in DCP-o-matic 2.14.x with 8 audio channels - 5.1 + VI (audio description).
Process: Rewrap to add closed captions in 2.16.59. Set output audio as 8 track. Disable HI track in audio matrix.
Output: SMPTE DCP with 16 audio channels.

The anomaly is that the MainSoundConfiguration node of CompositionMetadataAsset in the CPL reads 71/L,R,C,LFE,Lss,Rss,-,VIN,-,-,-,-,-,-,-,-.

The 71 part here seems inaccurate. Shouldn't it be 51?

I've not read the spec relating to this field so not 100% certain of the correct value, but have seen other commercially-produced DCPs with same audio layout where MainSoundConfiguration is listed as 51/L,R,C,LFE,Lss,Rss,-,VIN,-,-,-,-,-,-,-,-.

Please can you confirm what correct behavior is? If this is a bug, happy to provide a repro case if helpful.

TagsNo tags attached.
Estimated weeks required
Estimated work requiredUndecided



2023-06-30 18:55

administrator   ~0005799

Last edited: 2023-06-30 18:56

We did some tests with EasyDCP and a 5.1 DCP with HI + VI and a `MainSoundConfiguration of 71/L,R,C,LFE,Lss,Rss,HI,VIN and it passed. Doesn't mean that it's right though!


2023-06-30 18:58

administrator   ~0005800

It does look suspicious though, you're right.


2023-06-30 19:47

reporter   ~0005801

A few random data points... Here are some examples of DCPs we've been sent for festivals which use 51/:

DCP made by Deluxe with Cipher CplGen 1.8.23 (8 channels):

DCP made by Eikon with wl-cpl + Pylua v3.6.6 (8 channels):

DCP made by Motion Picture Solutions with Packman 3.6.3 (6 channels):

DCP made with Fraunhofer IIS easyDCP Creator+ 4.1.2 (16 channels):

DCP made with Fraunhofer IIS easyDCP Creator+ 4.1.2 (6 channels):

The first 2 are Netflix releases and the 3rd is Universal. That doesn't necessarily mean they're right, but I guess more likely to be.

I've searched our database and we haven't seen any DCPs at all in past year tagged 71/ except for ones made with DCP-o-matic.


2023-06-30 19:56

reporter   ~0005802

Actually, come to mention it, DCP-o-matic DCPs use Lss and Rss whereas all other DCPs from other software use Ls and Rs.

Contrary to what I said in first message in this issue, we've never seen 51/L,R,C,LFE,Lss,Rss, only 51/L,R,C,LFE,Ls,Rs.

We haven't received any 7.1 DCPs with SMPTE metadata so unable to advise what channel naming is used in those.


2023-06-30 22:50

administrator   ~0005803

The Lss/Ls difference should be between 5.1 and 7.1 (5.1 should use Ls, 7.1 Lss). Thanks for the references - they sound convincing!


2023-07-02 10:17

administrator   ~0005811



2023-07-02 11:03

reporter   ~0005813

Thanks Carl.

What will be written if the DCP has a different audio layout e.g.:

  • Mono (C channel only)
  • Stereo (L+R only)
  • 3.1 (L, R, C, LFE only)
  • 3.0 (L, R, C only)
  • No audio ("MOS")

I guess other layouts are theoretically possible, but the above are all seen "in the wild" fairly regularly.

The 2 really common ones are stereo (very common in shorts), and mono (not unusual in old "archive" films).

From commit 5e5a45917066e23be7ab8237c8dd63b21a69784e, it looks like only 5.1 and 7.1 are options.


2023-07-06 17:33

reporter   ~0005849

One other point (let me know if this should go in a separate issue):

When making a VF DCP to accompany an existing OV DCP, MainSoundConfiguration in the VF can be misleading if the OV doesn't have 16 channels.

e.g.: OV DCP has 6 audio channels only (5.1).
When making a VF, VF has MainSoundConfiguration: 51/L,R,C,LFE,Ls,Rs,-,-,-,-,-,-,-,-,-,-.

I believe this should reflect the number of actual channels in the DCP, so just 51/L,R,C,LFE,Ls,Rs (without the trailing ,-s).


2023-09-07 21:39

administrator   ~0005935

I think the excessive channels in VFs is fixed now. The documentation about this whole thing seems very vague but it seems like only 51/... and 71/... are allowed. Do you have any counterexamples from "reputable" sources?

Bug History

Date Modified Username Field Change
2023-06-30 16:35 overlookmotel New Bug
2023-06-30 18:55 carl Note Added: 0005799
2023-06-30 18:56 carl Note Edited: 0005799
2023-06-30 18:58 carl Note Added: 0005800
2023-06-30 19:47 overlookmotel Note Added: 0005801
2023-06-30 19:56 overlookmotel Note Added: 0005802
2023-06-30 22:50 carl Note Added: 0005803
2023-06-30 22:50 carl Assigned To => carl
2023-06-30 22:50 carl Status new => confirmed
2023-06-30 22:52 carl Branch => 2580-msc
2023-06-30 22:52 carl Estimated work required => Undecided
2023-06-30 23:26 carl Priority normal => urgent
2023-06-30 23:26 carl Severity minor => major
2023-07-02 10:13 carl Tag Attached: git-bug
2023-07-02 10:16 carl Tag Detached: git-bug
2023-07-02 10:17 carl Status confirmed => resolved
2023-07-02 10:17 carl Resolution open => fixed
2023-07-02 10:17 carl Note Added: 0005811
2023-07-02 11:03 overlookmotel Note Added: 0005813
2023-07-06 17:33 overlookmotel Note Added: 0005849
2023-07-08 15:44 carl Status resolved => acknowledged
2023-07-08 15:46 carl Resolution fixed => open
2023-08-13 12:23 carl Target Version => 2.16.x
2023-09-01 15:56 carl Target Version 2.16.x => 2.16.64
2023-09-07 21:39 carl Status acknowledged => feedback
2023-09-07 21:39 carl Note Added: 0005935
2023-09-13 22:34 carl Target Version 2.16.64 => 2.16.65
2023-09-21 00:08 carl Target Version 2.16.65 => 2.16.66